SunPump Meme Token Toolkit
Pass
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill utilizes the
@sun-protocol/sun-clipackage, which is the official command-line interface for the SUN.IO ecosystem. This is a well-known service for decentralized finance on the TRON blockchain. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted metadata from the TRON blockchain, such as token names, symbols, and descriptions, which could be used for indirect prompt injection attacks.
- Ingestion points: Blockchain data retrieved through the
sun sunpump token get,ranking, andsearchsubcommands specified inSKILL.md. - Boundary markers: Present. The skill includes an extensive "Agent Pre-Validation Checklist" and "Security Rules" requiring explicit user confirmation, previews of all write operations, and dry-run simulations.
- Capability inventory: Transaction broadcasting (launching, buying, and selling tokens) and reading local files for token logo uploads.
- Sanitization: Not explicitly provided in the instruction logic; the skill relies on the agent's reasoning capabilities and mandatory human-in-the-loop confirmation steps.
- [CREDENTIALS_UNSAFE]: The documentation references sensitive environment variables for wallet access (
TRON_PRIVATE_KEY,TRON_MNEMONIC). It adheres to security best practices by utilizing placeholders and providing clear, mandatory instructions for the agent to never request or display these values.
Audit Metadata